Compress PDF
Compress PDF reduces file size by re-encoding embedded images as JPEG at a quality you choose and by removing metadata and unused objects. Scanned and image-heavy documents shrink the most. A PDF that is mostly vector text is already compact, so savings there are small - the tool tells you honestly when that happens.
1 Add your file
Enter the open password to use it here. It is checked in your browser and never sent anywhere.
2 Choose your options
Compression works by re-encoding the images inside your PDF and dropping metadata. A scanned document, which is really a stack of pictures, can shrink by 60-90 per cent. A PDF that is mostly vector text is already about as small as it gets, and may barely change — the tool will tell you honestly if that happens rather than pretending otherwise. The strong setting turns pages into images, which shrinks aggressively but makes the text unselectable.
How to compress PDF
Add your PDF
Drop it in. We show its current size so you can judge the saving.
Pick a compression level
Recommended suits almost everything. Push to Smallest file for email attachments, or Flatten to images when nothing else has worked.
Compare and download
You see the before and after sizes and the percentage saved before you commit to the download.
Where the bytes actually are
PDF size is almost never spread evenly. It concentrates in images — a single phone photo can be 4 MB, and a scanned document is nothing but images, one per page. Embedded fonts come second, which matters mainly in short documents. Text content itself is tiny: a hundred pages of prose is often under 500 KB.
A quick diagnostic before you change any setting: divide the file size by the page count. Under roughly 100 KB per page and the document is mostly text, so there is little to win. Over 500 KB per page and it is images, which is where the savings live.
A worked example
A 40-page site report is 96 MB — about 2.4 MB per page, so images. The photographs were placed straight from a phone at roughly 500 DPI relative to the page, while a screen shows about 100 and a good printer uses 300. Downsampling to 150 DPI at quality 75 typically brings that file under 10 MB with no visible change on screen, because three-quarters of the pixels were never being displayed.
Limitations worth knowing
- Text-only and vector-heavy documents barely shrink. Content streams are already compressed, so there is nothing left to remove.
- Re-encoding at a higher quality than the original can make a file larger. Compression is not always subtraction.
- JPEG quality settings smear hard edges, so screenshots, diagrams and scanned text degrade visibly long before photographs do. Save size through resolution instead.
- Quality lost in compression cannot be recovered from the compressed copy — keep the original.
- Mail systems apply their size limit after encoding, which inflates a file by roughly a third. Aim for about 15 MB against a 20 MB limit.
Further reading
Related tools
Compress PDF — frequently asked questions
Because there was little to squeeze. Compression re-encodes images, and a PDF of vector text and shapes contains almost none. Files exported from Word or LaTeX are usually already compact. A scanned document, by contrast, is a stack of photographs and shrinks dramatically.
Yes, on every level except Flatten to images. The first three levels leave the text layer completely untouched and only rework the pictures. Flatten deliberately turns pages into images, and says so.
At the Recommended level most people cannot see a difference on screen. Smallest file becomes visible on photographs and on gradients. Line art and screenshots suffer more from downsampling than photographs do, so raise the DPI if you are compressing diagrams.
The tool accepts up to 100 MB, and warns above 50 MB. Beyond that the browser tab tends to run out of memory. Split the file first and compress the parts if you need to go bigger.
With Strip metadata ticked, the title, author, keywords and producing software are removed from the document properties. That is not the same as redaction — text inside the pages is untouched. Use Redact PDF for content.
Yes. Scans respond best of all. Try Recommended first, then Smallest file. If it is still too large, run OCR PDF first and then Flatten to images at a low DPI.